CITATION

Timmermans, B., C. Patricola, and M. Wehner. 2018. Simulation and analysis of hurricane-driven

extreme wave climate under two ocean warming scenarios. Oceanography 31(2):88–99,

https://doi.org/10.5670/oceanog.2018.218.

DOI

https://doi.org/10.5670/oceanog.2018.218

PERMISSIONS

Oceanography (ISSN 1042-8275) is published by The Oceanography Society, 1 Research Court,

Suite 450, Rockville, MD 20850 USA. ©2018 The Oceanography Society, Inc. Permission is

granted for individuals to read, download, copy, distribute, print, search, and link to the full texts of

Oceanography articles. Figures, tables, and short quotes from the magazine may be republished

in scientific books and journals, on websites, and in PhD dissertations at no charge, but the materi-

als must be cited appropriately (e.g., authors, Oceanography, volume number, issue number, page

number[s], figure number[s], and DOI for the article).

Republication, systemic reproduction, or collective redistribution of any material in

Oceanography is permitted only with the approval of The Oceanography Society. Please contact

Jennifer Ramarui at info@tos.org.

Permission is granted to authors to post their final pdfs, provided by Oceanography, on their

personal or institutional websites, to deposit those files in their institutional archives, and to share

the pdfs on open-access research sharing sites such as ResearchGate and Academia.edu.

OceanographyTHE OFFICIAL MAGAZINE OF THE OCEANOGRAPHY SOCIETY

DOWNLOADED FROM HTTPS://TOS.ORG/OCEANOGRAPHY

Oceanography | Vol.31, No.288 Oceanography | Vol.31, No.288

SPECIAL ISSUE ON OCEAN WARMING

SIMULATION AND ANALYSIS OF

HURRICANE-DRIVENEXTREME WAVE CLIMATE

UNDER TWO OCEAN WARMING SCENARIOS

By Ben Timmermans, Christina Patricola, and Michael Wehner

Oceanography | June 2018 89

INTRODUCTIONOcean waves driven by wind stress—wind-waves—are familiar and often impressive, bringing the benefits of renewable energy and recreational activ-ities like surfing, but frequently acting as hazards to coastal and offshore industries. For example, in recent years extreme lev-els of erosion on coasts of the European Atlantic (Masselink et  al., 2016) and US Pacific (Barnard et  al., 2017), attrib-utable to extremely energetic wave condi-tions, have been reported. Barnard et al. (2017) showed that wave energy inci-dent to the US Pacific coast, linked to the strong El Niño of 2015/2016, contrib-uted to the largest observed integrated wave energy based on historical records. However, possible changes in wave inten-sity are not always the most import-ant finding because waves are complex and multivariate phenomena. Harley et  al. (2017) attributed the most severe coastal erosion observed in southeast Australia in 40 years to the joint effect of wave height and (anomalous) wave direc-tion. With changing polar climate, there has been substantial interest in wave-sea ice interaction (Doble and Bidlot, 2013; Casas-Prat et al., 2018), with wave action linked directly to Antarctic ice-shelf collapse (Massom et  al., 2018). Island nations in particular are susceptible

to changes in wave conditions (Hoeke et  al., 2015; Duvat et  al., 2016), and such severe impacts, noted in particu-lar by the Intergovernmental Panel on Climate Change (IPCC; IPCC, 2014), motivate efforts such as the Coordinated Wave Climate Intercomparison Project (COWCLIP; Hemer et  al., 2012, 2013a; X.L. Wang et  al., 2016) that is in part supported through the submission of simulated global and regional wave climate data sets, two of which are described in this paper.

With a strong research imperative, studies have focused on both historical and future conditions in various ways. Satellite and data buoy observations cap-ture both the climate (Ruggiero et  al., 2010; Izaguirre et al., 2011; Young et al., 2012) and specific events (D.W. Wang et  al., 2005), while reanalysis data have been analyzed using dynamical and sta-tistical methods (Camus et  al., 2014). Dynamical projections typically involve the generation of near-surface winds using a (possibly coupled) atmosphere model that can then be used to derive ocean waves either statistically (X.L. Wang and Swail, 2006; Perez et al., 2015; Camus et al., 2017) or dynamically (Hemer et al., 2013b) using a global wave model such as WAVEWATCH III (WW3; Tolman, 2014). So-called “phase-averaged” global

models such as WW3 employ an aver-aged spectral representation of waves and do not resolve individual waves (Komen et  al., 1994), including extreme “rogue waves,” although they can be informative (Babanin and Rogers, 2014). WW3 has been used widely for global (Timmermans et  al., 2017) and regional studies (Shimura et  al., 2016; Aguirre et  al., 2017), and wave energy resource assessment (Mackay et al., 2010; Hemer et al., 2017).

Research suggests that wave climate may change substantially in the future, with considerable regional variation (Hemer et al., 2013a; Erikson et al., 2015; Shimura et  al., 2016; Aarnes et  al. 2017; Casas-Prat et  al., 2018). Erikson et  al. (2015) evaluate wave climate in the east-ern north Pacific and report changes in both mean and extreme wave conditions, driven predominantly by extratropical cyclones (ETC) in the mid-latitudes, with a generally decreasing trend with sever-ity of greenhouse gas forcing. Swell gen-erated by ETCs in the Southern Ocean—consistently the roughest on Earth (Young 1999)—is likely to result in future increases in wave height (Hemer et  al., 2013a) that may contribute to inunda-tion threat to islands in the western trop-ical Pacific (Shope et al., 2016). Shimura et al. (2016) found that winter wave con-ditions in the western North Pacific were dominated by ETCs with inten-sity set to decrease under future scenar-ios. They remark in particular that local coastal conditions, and resultant impacts, in Japan are a complex function of wind-sea and remote swell and advocate for focused high-resolution regional studies in such areas in order to attribute effects.

Extreme wave conditions that typi-cally present the greatest risks are also set to change (Fan et al., 2013; X.L. Wang et al., 2014; Barnard et al., 2017; Shimura et al., 2015, 2016, 2017), although inves-tigation of extremes can be substantially more challenging than assessment of means. Inference about extremes (Coles, 2001; see also section on Extreme Value Analysis) requires long time series of

ABSTRACT. Ocean wave climate is an important area of research, particularly in the context of extremes driven by tropical cyclones (TC). We can now simulate global cli-mate at resolutions sufficient to resolve TCs and for durations long enough to explore climatological changes. Both the devastating 2017 North Atlantic hurricane season and growing evidence for the connection between TC activity and increasing ocean tem-perature motivate investigation of possible future changes. We present two simulated 50-year global wave climate data sets under possible future warming scenarios char-acterized by +1.5°C and +2.0°C stabilized global mean temperatures that capture the effects of TCs. Differences in extreme wave climate between these possible scenarios and present-day conditions appear to be significant in many areas, particularly those affected by TCs. However, for computational feasibility, simulations of this kind rely on fixed sea surface temperatures, so we also investigate and elucidate effects from the lack of a dynamic ocean by simulating waves from a number of recent hurricanes and comparing output to observations. We conclude that atmosphere-only forcing is likely to result in an overestimate of extreme wind speeds and wave heights in TC-affected regions. More ensemble studies are needed to help elucidate detailed processes relevant to extreme wave climate, and important community projects such as the Coordinated Wave Climate Intercomparison Project (COWCLIP) should be supported.

Oceanography | Vol.31, No.290

output to robustly quantify the probabil-ity of rare events. Furthermore, a model may be limited in its ability to accurately reproduce physical processes respon-sible for extremes. Many of the afore-mentioned studies are concerned with impacts from waves driven by ETCs—large synoptic-scale storms—that typi-

cally originate in the mid-latitudes and impact the coasts of northern Europe and the northwestern United States. Being spatially broad, sometimes covering thou-sands of kilometers, they allow waves to develop over very long fetch. Numerical investigation can be very effective for storms of this kind, which are typically well resolved at horizontal resolutions around 1° (~100 km). Such simulations are affordable to run and output is abun-dant (Taylor et al., 2012). In contrast, trop-ical cyclones (TC) are not well resolved at spatial and temporal resolutions typically employed in global atmosphere simula-tions (Timmermans et al., 2017), but they are clearly important to atmospheric and oceanic climate and impacts (Sriver, 2016; S.S. Chen and Curcic, 2016).

While damage from hurricanes (the strongest TCs) is the dominant economic cost from “billion dollar” natural disas-ters (~63% of total) in the United States (NOAA NCEI, 2018), 2017 in particular was a stark reminder of the threat of hur-ricanes to society. Three major hurricanes made US landfall, setting records, includ-ing the most rainfall in a single storm (Harvey) and the largest annual losses due to hurricanes in US history (Munich RE, 2018). Hurricane costs attributable to waves are difficult to ascertain because

wind, waves, and storm surge act in com-bination, but they can be substantial. During Katrina, 1,500 people lost their lives on shore due to storm-surge-related flooding, while offshore, dozens of oil and gas platforms were damaged or destroyed, resulting in near total shutdown of the Gulf of Mexico’s offshore production.

Research on TC climatology and related effects at the atmosphere-ocean boundary, including waves, is challeng-ing. TCs and hurricanes form from atmo-spheric disturbances typically at least 5° north or south of the equator. When developed, the structure of the storm’s center—the eye—is typically smaller than ~100 km and thus, where numerical sim-ulation is required, resolution needs to be significantly higher than this in order to even approximate the process. High-resolution coupled modeling, required to resolve various complex effects (Zambon et al., 2014; S.S. Chen and Curcic, 2016; Fan and Rogers, 2016), remains prohib-itively expensive for climatological study. Nonetheless, it turns out that atmo-spheric simulations below ~60 km hor-izontal resolution can produce realistic TC climatology (Murakami et  al., 2012, 2015). Furthermore, use of fixed sea sur-face temperature (SST) boundary condi-tions (Hurrell et  al., 2008), in place of a dynamic ocean, reduces computational cost, rendering this kind of approximate approach feasible for well-resourced institutions. Resulting winds, including TCs, can be used to drive a wave model, although such studies remain sparse (Fan et  al., 2013; Shimura et  al., 2015; Timmermans et al., 2017).

In this paper, using such a numerical approach, we augment the investigation of Timmermans et al. (2017) by present-ing two future wave climate data sets and an analysis of changes in the extremes, in terms of the 20-year significant wave height, Hs, return level. The two scenarios follow the Half a degree Additional warm-ing, Prognosis and Projected Impacts (HAPPI) protocol (Mitchell et al., 2017), representing 1.5°C and 2.0°C stabilized climates agreed upon as long-term targets to mitigate severe climate change impacts by members of the United Nations as part of the 2016 Paris agreement. However, cognizant of possible uncertainties intro-duced through this approach, and dis-cussed further in the next section, we also make use of recent hindcasts of a num-ber of major US hurricanes. This allows us to compare directly with observations from data buoys and investigate sources and magnitudes of error that may exist in approximate climatological simulations. We find a tendency toward excessively extreme Hs, but with accuracy in more benign conditions, which raises ques-tions about the validity of simulations of (extreme) wave climate and motivates fur-ther research and corrective strategies. In the following section, we elaborate further on our understanding of TC and wave cli-mate, and investigate it through numer-ical approaches. Our approach to wave modeling, and subsequent application of extreme value analysis are described next. We then discuss results of the analysis of future extreme wave climate and describe the hurricane wave simulations and their analysis. We briefly offer our conclusions in the final section.

TROPICAL CYCLONES AND WAVESEnergetically, TCs can be considered to be a Carnot heat engine, powered by energy from warm ocean surface waters. The dependence of TC gene-sis and evolution on SST has been stud-ied at length (Emanuel, 2005; Vecchi and Soden, 2007a), although properties of the atmosphere such as the temperature

“Both the devastating 2017 North Atlantic hurricane season and growing evidence for the connection between tropical cyclone activity and increasing ocean temperature motivate investigation of possible future changes.

”.

Oceanography | June 2018 91

of the tropopause (Emanuel et al., 2013) and vertical wind shear (Vecchi and Soden, 2007b) are particularly influ-ential. Feedback from the ocean is also important: mixing of cooler subsurface water by induced turbulence reduces heat availability and results in less- favorable conditions. Seroka et  al. (2016) show that during Hurricane Irene (2011), “ahead-of-eye” cooling of coastal sur-face waters due to currents induced by Irene itself, was a key causal factor in rapid de-intensification before mak-ing landfall. Conversely, the availabil-ity of warmer subsurface water may sup-port particularly powerful events, as speculated for “Super Typhoon” Haiyan in 2013 (Lin et  al., 2014). A steepening subsurface temperature gradient under global warming may give rise to a greater cooling effect, thus suppressing intensi-fication in future climates (Huang et al., 2015). TC activity is also linked to the El Niño-Southern Oscillation (Patricola et  al., 2014, 2016; Zhang et  al. 2016), but as Murakami et al. (2017) point out, many factors are influential, and a strong El Niño does not solely predicate an active hurricane season. In general, the question of how climate change is affect-ing TC climatology remains the subject of debate (Sobel et al., 2016).

Ocean waves arise from momentum transfer through wind stress—a func-tion of surface drag—itself affected by the waves (Komen et  al., 1994; Moon et  al., 2003). However, complex atmosphere- ocean- wave interactions, particularly under extreme conditions, remain poorly understood. Y. Chen and Yu (2017) found that choice of wind stress parameteriza-tion resulted in differences of up to 12 m in peak wave heights under hurricane conditions. Research has also examined turbulent mixing by waves (Aijaz et  al., 2017) and the effect of induced currents on wave height around the eye of the hur-ricane (Fan et  al., 2009). Investigation of these complex interactions typically requires high (~ few kilometers) resolu-tion coupled simulations (Zambon et al., 2014; S.S. Chen and Curcic, 2016), but in

general, the use of coupling in climato-logical simulations remains limited. For example, coupling between atmosphere- wave (Fan et  al., 2013) and atmosphere- ocean (Li and Sriver, 2018) is possi-ble on short (~100 year) climatological timescales. However, Fan et  al. (2013) report that the high-resolution atmo-spheric model (HiRAM; ~ 50 km resolu-tion) generates few major hurricanes. In addition, although the Community Earth System Model (CESM; ~25 km resolution) has been shown to produce remarkably good TC intensity distributions, includ-ing major hurricanes, when using fixed SSTs (Wehner et al., 2015), Li and Sriver (2018) show that coupling the atmosphere to both slab and dynamic ocean alleviates spatial bias in TC storm track distribu-tion seen in the uncoupled case. In gen-eral, however, tropical SST biases com-mon to generations of coupled climate models (Zuidema et al., 2016) are known to cause substantial errors in simulated TC activity, with an under-simulation of 50% in the Atlantic and over- simulation of 80% in the east Pacific (Wei-Ching Hsu, Texas A&M University, pers. comm., 2018). Noting the unavoidable uncer-tainties, and difficulty in finding a tracta-ble approach, the computational advan-tage of using fixed SSTs is attractive, and this configuration has been used widely. Mizuta et  al. (2017), for example, ran 5,000 years of global atmospheric simula-tions at 60 km horizontal resolution, suf-ficient to resolve the TC frequency distri-bution, although statistical bias correction methods are required to investigate the most intense TCs, because they cannot be resolved numerically. Shimura et  al. (2015) used some of those simulations to examine extreme wave heights from TCs in the western North Pacific, alleviating the conditionality on fixed SSTs by using an ensemble. The analysis of extreme wave climate in this study (see section on Changes in Extreme Wave Climate) therefore utilizes atmospheric simula-tions bounded by fixed SSTs, acknowl-edging uncertainties associated with the approach described above.

APPROACH TO WAVE MODELINGThe approach to wave climate model-ing and analysis follows Timmermans et  al. (2017), briefly summarized here. In the later section on Hurricane Wave Modeling, we also make use of 10 m winds from high-resolution (3 km, 4.5 km, and 27 km) simulations drawn from ensem-bles of individual hurricanes employing fixed (observed) SSTs generated by the Weather Research and Forecasting Model (WRF; Skamarock et al., 2008). For these, the wave model configuration is simi-lar to that described here but employs regional grids (Gulf of Mexico and trop-ical western Atlantic) at similar resolu-tions (4 km and 27 km) and appropriate integration time steps. Note that in those cases, focusing on the locally generated hurricane waves, sea surface boundar-ies were left open, thus ignoring remotely generated wave systems. More discussion of the WRF simulations is provided in the section on Hurricane Wave Modeling.

For global climate, 10 m three-hourly winds at ~25 km horizontal resolution were obtained from simulations of the atmosphere using the CESM (Hurrell et  al., 2013) bounded by SST patterns commensurate with +1.5°C and +2.0°C stabilized climates. Specifications of SST patterns are given by Mitchell et  al. (2017; their section 2.1) but, to sum-marize construction of the +1.5°C sce-nario, the difference between the aver-age of 2090–2100, taken from Coupled Model Intercomparison Project Phase 5 (CMIP5) RCP2.6 scenario ensemble, and the average of 2006–2015, is added to observed SSTs for 2006–2015. Five inde-pendent CESM simulations, each initial-ized with a microscopically perturbed atmospheric state, generate winds for the 10-year period “2106–2115,” which are then used to force WW3. For each sce-nario, we make the assumption that each year of output is an independent sample in a stationary climate, thus providing 50 years for extremal analysis. WW3 is configured on a 0.25° global grid (corre-sponding approximately to the resolution of the forcing winds), utilizing ST4 input

Oceanography | Vol.31, No.292

and dissipation scheme and NL1 (DIA) nonlinear interactions (see Timmermans et al., 2017, for more details), with three-hourly output of global fields of various wave parameters. Hs at each model grid cell is analyzed using an extreme value approach, outlined in the next section. Note also that in the section Changes in Extreme Wave Climate, we compare out-put in terms of 20-year return level to output from a similar 44-year data set representative of present-day climate, described in Timmermans et al. (2017).

EXTREME VALUE ANALYSISExtreme value (EV) theory is an active area of statistical research (Coles, 2001; Davison and Huser, 2015) and of ever- increasing importance in environmen-tal studies, as interest in future extreme events grows (Risser and Wehner, 2017). It provides a statistical framework for the study of rare events and allows us to esti-mate, for example, return levels and peri-ods of wave height. The use of univari-ate EV theory is now fairly common and facilitated by many off-the-shelf pack-ages, such as the extRemes (http://www.assessment.ucar.edu/toolkit/) package for R (https://www.r-project.org/). Méndez et al. (2006), for example, make use of EV theory to investigate trends in extreme Hs in the northeastern Pacific. However, many questions exist about more complex multivariate extreme phenomena where

methodological challenges remain sub-stantial (Davison and Huser, 2015). Here, following Timmermans et al. (2017), we make use of the “peaks- over- threshold” (POT) method (Coles, 2001), which involves statistically modeling threshold exceedances of Hs (typically >97.5 per-centile), in order to derive an EV distri-bution (EVD) from which 20-year return levels can be derived. Note that estimates of decay rate in the tail of the probability distribution can be sensitive to the occur-rence of a single high-magnitude event so here, noting the high spatial variabil-ity of extreme waves, we actually employ a dynamic threshold algorithm to fit the EV model more robustly.

For brevity, we omit further details of fitting an EVD and refer the reader to Timmermans et  al. (2017), but we illus-trate some important issues by examining observed and simulated wave height data. Figure 1 panels a and b show histograms for Hs at two NOAA data buoys that lay in the paths of hurricanes Irma and Maria, respectively. The general shape of the distributions is typical of the open ocean and is skewed, with a right-hand tail due to higher winds. Note that buoy 41047 (Figure 1b), showing a slower tail decay rate, lies in the North Atlantic and is exposed to more energetic conditions, in contrast to buoy 42060 (Figure 1a) that is more sheltered in the Caribbean (buoy locations are shown in Figure 3,

bottom left panel). However, the distribu-tion statistics tell us very little about the most intense events, which are so infre-quent that they cannot be resolved in the figures—maximum, pre- and post- hurricane passages (blue and red arrows, respectively) are indicated.

An EVD could be fitted using a POT approach to model the occurrence of the extremes; however, in these cases we can see that in approximately 10 years of observations, the variability in the max-imum is substantial—a single event can result in an increase in the maximum of between 25% and 100%. In fact, at buoy 42060, the single event of hurricane Irma is responsible for all observations in the highest 50% of the range of the data, so we have only a single independent data point (high threshold exceedance) from which to infer extremal behavior. Figure 1a includes point estimates and 95% con-fidence intervals, based on the statisti-cal sampling only, for 10-year return lev-els for Hs. The estimates were derived from the data excluding (light blue) and including (orange) the passage of Irma, revealing that the earlier 95% confidence interval does not bound wave heights from the passage of Irma, and in fact sug-gests them to be extremely unlikely. In this case, the short duration of the obser-vational record with respect to the fre-quency of TC passage (and extreme waves) in this region limits our ability

FIGURE 1. Histograms for significant wave height (Hs) at NOAA data buoys (a) 42060 (Central Caribbean Sea) and (b) 41047 (Hatteras Plain, Western Atlantic). Arrows indicate the maximum value before (blue) and after (red) hurricane passage. Panel (c) shows 50 years of simulation output, with the (red) arrow indicating the maximum value.

(a) Irma: NDBC 42060 (8 years)

Hs (m)

Freq

uenc

y

0 2 4 6 8 10 12

050

0010

000

1500

0

� �

10−yr return level estimatespre− (blue) & post− (orange)

hurricane passage

(b) Maria: NDBC 41047 (11 years)

Hs (m)

Freq

uenc

y

0 2 4 6 8 10 12

050

0010

000

1500

0

(c) Arabian Gulf [62°E,15°N] (50 years simulated)

Hs (m)

Freq

uenc

y

0 5 10 15 20

010

000

3000

050

000

Observed and simulated significant wave height statistics

Oceanography | June 2018 93

to infer extremal behavior. For compari-son, Figure 1c shows an example of a his-togram of Hs from simulated wave cli-mate (our +2.0°C data set of 50 years) at a location in the Arabian Gulf affected by TCs on an infrequent basis. Note that the maximum value of approximately 22.5 m is indeed extreme, exceeding the largest in the observed record of approximately 19 m, which was in the North Atlantic and not due to a TC.

CHANGES IN EXTREME WAVE CLIMATEWe fitted independent EVDs to each grid cell for each of the 50-year wave climate data sets and obtained 20-year return levels for Hs. We compare these to a similar (44-year) data set described by Timmermans et  al. (2017) in order to evaluate changes. Figure 2 shows differences in 20-year return levels for Hs between future +1.5°C and the

present day (panel  a), +2.0°C and the present day (panel  b), and +2.0°C and +1.5°C (panel  c). In terms of magni-tude of change, Figure 2 panels a and b look remarkably similar to each other, and also to Figure 3c of Timmermans et al. (2017), which shows the difference between the present day and the future RCP8.5 scenario. These results indicate only small differences in extreme wave climate between the future scenarios.

FIGURE 2. Differences in Hs 20-year return level between (a) +1.5°C and the present day, (b) +2.0°C and the present day; (c) +2.0°C and +1.5°C. Stippling indicates areas where the difference is estimated to be statisti-cally significant.

a

b

c

Oceanography | Vol.31, No.294

However, there may be evidence of some large regional changes, of many meters, between present-day and future scenar-ios. A higher density of storm tracks in the central and eastern Pacific is seen in all future scenarios, with possible evi-dence of a reduction in extreme waves (due to reduced TC activity) in the Pacific east of Australia.

These possible regional changes in extreme wave climate reflect the simu-lated distribution of storm tracks in the

warmer scenarios, which raises questions about the shift. Li and Sriver (2018, their Figure 5a–h) show the difference between TC climatologies arising from fixed SSTs and a coupled ocean when simulated in CESM. With respect to observations and coupled modeling, fixed SSTs lead to a high bias in the density of TCs, par-ticularly in the eastern Pacific and the Atlantic and Indian Oceans. Here, how-ever, the easterly bias in TC track den-sity associated with fixed SST simulations

appears to be further compounded in the projected climate. In the South Pacific, east of Australia, where track density bias appears to be lower, a possible reduc-tion in extreme wave heights is apparent. However, owing to the limited duration of the data sets, the extremes are substan-tially affected by individual TC tracks, and robust determination of differences is challenging. This is particularly prob-lematic in areas of low frequency of pas-sage of intense TCs. A particular example can be seen in Figure 2b in the Arabian Gulf, where a single event—bearing resemblance to Cyclone Gonu (2007)—dramatically affects the 20-year Hs return level (see also Figure 1c). Note that dark stippling provides an estimate of statis-tical significance of changes (calculated from estimates of the sampling distribu-tion when fitting the EVD), revealing that robustness of difference is mostly lim-ited to extremes arising from individ-ual storms, including ETCs in the mid- latitudes. This challenge motivates both the generation of longer duration data sets and further detailed analysis of exist-ing data (e.g., Shimura et al., 2017).

HURRICANE WAVE MODELINGThe large Hs apparent in our data sets (max. 22.5 m), and reported in Timmermans et al. (2017), may be com-mensurate with excessive TC intensity in simulations using fixed SSTs, as suggested by Li and Sriver (2018). We elucidate this for waves from hindcasts of individual hurricanes, generated using WRF, with a fixed SST surface boundary condition. Although WRF and CESM differ in for-mulation and application, given the com-mon lack of coupling, we anticipate errors in WRF-driven simulations to be infor-mative with respect to potential errors in CESM. We have access to ensemble simu-lations of a range of hurricanes, designed to robustly detect response to increased SSTs (Patricola and Wehner, in press). These simulations employ local fixed- resolution grids of 3 km (for Katrina) and 4.5 km, and are forced by observed SSTs and lateral boundary conditions

Katrina

4200142003

42036

4203942040

24

26

28

30

32

34

−90 −85 −80 −75

Historical 27 kmHistorical 3 kmObservedRCP8.5 3 km

Ike

4200142002

42035

4203942040

42055

20.0

22.5

25.0

27.5

30.0

−95 −90 −85 −80 −75

Historical 27 kmHistorical 4.5 kmObservedRCP8.5 4.5 km

Irma

42060

41043

41046

42036 41009

15

20

25

30

−80 −70 −60

Historical 4.5 kmObservedRCP8.5 4.5 km

Maria

42060

41043

41046

41047

15

20

25

30

−80 −70 −60 −50

Historical 4.5 kmObservedRCP8.5 4.5 km

Katrina

4200142003

42036

4203942040

24

26

28

30

32

34

−90 −85 −80 −75

Historical 27 kmHistorical 3 kmObservedRCP8.5 3 km

Ike

4200142002

42035

4203942040

42055

20.0

22.5

25.0

27.5

30.0

−95 −90 −85 −80 −75

Historical 27 kmHistorical 4.5 kmObservedRCP8.5 4.5 km

Irma

42060

41043

41046

42036 41009

15

20

25

30

−80 −70 −60

Historical 4.5 kmObservedRCP8.5 4.5 km

Maria

42060

41043

41046

41047

15

20

25

30

−80 −70 −60 −50

Historical 4.5 kmObservedRCP8.5 4.5 km

FIGURE 3. (left panels) Hurricane tracks for the different simulations (for each scenario) used and locations of relevant NOAA data buoys. (right panels) Corresponding maximum wind speeds.

Oceanography | June 2018 95

from NCAR’s Climate Forecast System Reanalysis (CFSR) data. Each ensemble is composed of 10 simulations for each of historic and RCP8.5 (observed +2.0°C elevated SSTs) scenarios. We studied four hurricanes—Katrina (2005, category 5), Ike (2009, category 4), Irma (2017, cate-gory 5), and Maria (2017, category 5)—and selected from each ensemble wind fields from the simulation that exhibited a storm track that corresponded most closely to observations. In the case of hur-ricanes Ike and Katrina, we also (fortu-itously) have access to a single simulation at 27 km, commensurate with the reso-lution used for our climatological sim-ulations. The output time step was set at 15 minutes. Wave fields were generated every 15 minutes using WW3, configured as described earlier in Approach to Wave Modeling. For these simulations in partic-ular, unbounded numerical grids at 4 km resolution were employed, spanning the Gulf of Mexico (Katrina and Ike) and a region extended to the east to capture the coastal Atlantic (Irma and Maria). While this potentially excludes incoming wave systems crossing the grid boundary, no significant events were identified during the periods of interest, and observations suggest that the hurricanes were solely responsible for the extremes in each case. Furthermore, initial condition error was judged to be minimal, given that initial Hs observations were relatively small (<1 m) in all cases (see, e.g.,  Figures 5 and 6), and discrepancy with simulation output was also typically small, indicating that spin-up occurred within a few hours.

Figure 3 shows storm tracks and max-imum wind speeds. While Hurricane Katrina principally impacted the Gulf of Mexico, the other three events had longer storm tracks spanning the outlying Caribbean islands. Figure 3 shows that the selected tracks, while similar, exhibit discrepancies from observation in space and time. Furthermore, the right-hand panels show discrepancies in the magni-tude and timing of the maximum wind speeds, compared with HURDAT2 obser-vations (http://www.aoml.noaa.gov/hrd/

hurdat/Data_Storm.html). Spatial dis-crepancies tend to be small compared with the radius of the hurricane, although the divergence is more significant for Hurricane Maria. These discrepancies affect subsequent simulations of waves, although the objective is not to perfectly recreate each hurricane but rather, by comparison with observations, to eluci-date sources of error and uncertainty in simulated extreme wave conditions.

We begin by comparing simulated 10  m wind speed (U10) and Hs with observations at a number of NOAA

data buoys, and in the interest of brev-ity we only show more detailed results from Hurricane Ike. Note that other hurricanes were qualitatively similar in terms of deviation from observations. Simulations were conducted at 4 km and 27 km using historical winds, and at 4 km using winds conditioned on an RCP8.5 scenario forced by warmer SSTs. During Ike, the eye passed very close to buoys 42001 and 42019, thus provid-ing an opportunity to examine the most intense part of the storm. Figure 4 shows comparisons of U10.

FIGURE 4. Hurricane Ike: Simulated 10 m wind speed (U10; top half of each panel, solid and dashed lines) under historical conditions at 4 km (blue), 27 km (light blue), and RCP8.5 winds, 4 km only (red) compared with observations from NOAA data buoys (circles and triangles), and wind direction (bot-tom half of each panel) simulated under historical conditions at 4 km (crosses) compared with obser-vations (squares).

(a) NDBC 42001

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(b) NDBC 42002

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(c) NDBC 42035

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(d) NDBC 42039

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(e) NDBC 42040

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(f) NDBC 42055

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(a) NDBC 42001

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(b) NDBC 42002

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(c) NDBC 42035

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(d) NDBC 42039

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(e) NDBC 42040

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(f) NDBC 42055

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(a) NDBC 42001

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(b) NDBC 42002

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(c) NDBC 42035

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(d) NDBC 42039

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(e) NDBC 42040

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

(f) NDBC 42055

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

010

2030

4050

U10

(ms)

Win

d D

ir. (d

eg)

090

180

360

U10 (Obs.)Wind Gust (Obs.)U10 (Hist. 4 km)U10 (Hist. 27 km)U10 (RCP 8.5 4 km)Wind Dir. (Obs.)Wind Dir. (Hist. 4 km)

Oceanography | Vol.31, No.296

The three simulations (historical, RCP8.5, and 27 km) reveal a range of agreement with observations, with sev-eral notable features. Considering the his-torical simulation (solid blue line) first, between Cuba and landfall in Texas, the simulated track coincided closely with the observed track. Agreement appears good with observations for both U10 and direction, with the exception of excessive U10 (>20 m s–1) at buoy 42001 in panel a. Buoy 42001 was fortuitously positioned in the path of Ike, and in fact, from both

observations and simulation, the passage of the eye is indicated by a rapid drop in wind speed, around 2008-09-12 00:00, before a subsequent rapid increase. The high wind speeds shown in Figure 4a are commensurate with the discrepancy between simulated and observed peak wind speed estimates from HURDAT2 data shown in Figure 3. In contrast, at distances further from the hurricane’s eye, U10 agreement is much better. The RCP8.5 (red line) and 27 km histori-cal (dashed blue line) cases show more

variability, with discrepancies in timing and intensity. In particular, peak inten-sity of the 27 km run leads the others by at least 12 hours, likely due to errors in translation speed, seen also between the historical and RCP8.5 simulations. High wind speeds and temporal lag at buoy 42002 appear to be due to divergence of the track (red line), which passes much closer to 42002 than the historic track (see Figure 3), and makes landfall some 200 km west of observations. We also note there is no clear (visual) evidence that the RCP8.5 simulation consistently yields higher wind speeds, although the 27 km simulation does appear to gen-erate 40% lower wind speeds at buoys 42001 and 42002. Note also that buoy 42035 is located in shallow water (15 m), where depth-induced breaking would be expected to constrain wave height in par-ticularly energetic conditions.

We find that the characteristics of the wind speed are reflected closely in the wave simulations, shown in Figure 5. We compare simulation output to buoy mea-surements for Hs (top half panel) and peak period, Tp, (4 km historical simula-tion only, bottom half panel). Tp appears to be consistently well reproduced, gen-erally falling within observational scatter (buoy observations typically lack uncer-tainty estimates). In general, discrepan-cies in Hs follow discrepancies in U10, with some variation. For example, his-toric U10 at buoy 42002 (solid blue line, Figure 4b) closely follows observations, but Hs exceeds observations by at least 2 m (solid blue line, Figure 5b). Discrepancies in U10 for the RCP8.5 and 27 km simu-lations also tend to be reflected in the Hs output. In general, large positive errors tend to occur close to the storm track, and much smaller, mostly negative, errors further away.

For comparison, Figure 6 shows simi-lar analysis of Hs and Tp at the buoy loca-tion(s) showing the most intense condi-tions across all four hurricanes. Note in particular that buoy 42003 (Figure 6b) failed during Katrina (the first loss of a NOAA deep water buoy in 30 years of

(a) NDBC 42001

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(b) NDBC 42002

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(c) NDBC 42035

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(d) NDBC 42039

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(e) NDBC 42040

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(f) NDBC 42055

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (Hist. 27 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

FIGURE 5. Hurricane Ike: Simulated Hs (top half of panel, solid and dashed lines) under histori-cal conditions at 4 km (blue), 27 km (light blue), and RCP8.5 winds, 4 km only, (red) compared with observations from NOAA data buoys (circles), and peak wave period (bottom half of panel) under historical conditions at 4 km (blue dashed line) compared with observations (crosses).

Oceanography | June 2018 97

operations in the Gulf), hence the loss of data, and that the second peak in Hs, seen in Figure 6e, is in fact hurricane Jose (but not relevant here). Four of the six exam-ples show maximum Hs that substantially exceed observations, while, similar to the case for Hurricane Ike, performance for Tp generally seems good in spite of numerous potential sources of uncer-tainty. Anomalous timing and magni-tude of some simulations can be related directly to storm tracks. For example, although the timing is well synchronized in the three cases for Katrina, discrepancy in wave heights at buoy 42001 (Figure 6a) appears to be explained by the proximity of the respective tracks to the buoy when intensity was highest. Given that the his-toric track (solid blue line, Figure 3) so closely matches the observed track, but simulated Hs exceeds observations by approximately 35%, it appears that high wind speeds are substantially biasing the extremes. This also seems likely at buoy 42060 during Hurricane Maria (panel f), where although simulated storm tracks appear realistic, Hs dramatically exceeds observations.

Notably, Y. Chen and Yu (2017) also raise the issue of high simulated Hs at buoy 42001 (see their Figure 6), during Katrina. They speculate that the hurri-cane’s rotation created offshore winds on its left side, thus resulting in fetch-limited conditions due to the coastline, which might be more poorly represented by the modeling setup. We assert that there may be more to this issue because our results appear to show consistent over-estimation of peak Hs in a number of cases where fetch is essentially unlimited. Indeed, Y. Chen et al. (2018) call for more expansive studies by looking at addi-tional hurricanes in order to draw more robust conclusions.

CONCLUSIONSWe presented two 50-year simulated global wave climate data sets under +1.5°C and +2.0°C warming scenar-ios. Changes in extreme wave height may be evident, particularly increases in

the tropical North Pacific and Atlantic basins, and possible reduction in the tropical South Pacific. However, these changes are characterized by poor sam-pling of TCs, which introduces uncer-tainty, and in turn motivates the need for much longer duration data sets. Furthermore, recent research suggests that the use of fixed SST boundary con-ditions in CESM may introduce regional bias in storm track density, so we advise caution in interpretation. Simulations of individual hurricanes using the WRF

model with observed SSTs suggest that extreme wind speeds, and resulting wave heights, are likely to be excessive. Noting also that evolution of simulated storm track, intensity, and translation speed contribute substantially to variability in output wind and wave characteristics, we echo Y. Chen et al. (2018) and advo-cate for broader ensemble studies of this type. Such data sets would add value to COWCLIP and help advance the investi-gation of the effects of atmosphere-ocean processes on extreme wave climate.

A) (a) Hurricane Katrina (2005): NDBC 42001

2005−08−23 2005−08−25 2005−08−27 2005−08−29 2005−08−31

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Hs (Hist. 27 km)Tp (Obs.)Tp (Hist. 4 km)

(b) Hurricane Katrina (2005): NDBC 42003

2005−08−23 2005−08−25 2005−08−27 2005−08−29 2005−08−31

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Hs (Hist. 27 km)Tp (Obs.)Tp (Hist. 4 km)

(c) Hurricane Ike (2008): NDBC 42001

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Hs (Hist. 27 km)Tp (Obs.)Tp (Hist. 4 km)

(d) Hurricane Ike (2008): NDBC 42035

2008−09−06 2008−09−08 2008−09−10 2008−09−12 2008−09−14

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Hs (Hist. 27 km)Tp (Obs.)Tp (Hist. 4 km)

(e) Hurricane Irma (2017): NDBC 41043

2017−09−04 2017−09−06 2017−09−08 2017−09−10 2017−09−12

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

(f) Hurricane Maria (2017): NDBC 42060

2017−09−17 2017−09−19 2017−09−21 2017−09−23 2017−09−25

02

46

810

Hs (m

)

T p (s

)5

1015

20

Hs (Obs.)Hs (Hist. 4 km)Hs (RCP 8.5 4 km)Tp (Obs.)Tp (Hist. 4 km)

FIGURE 6. Wave parameters at selected buoys where conditions were the most energetic due to Hurricanes Katrina, Ike, Irma, and Maria simulated with historical 4 km winds (blue), historical 27 km winds (light blue, where applicable), and RCP8.5 winds (red), compared with observations (circles and crosses) at buoys that observed the largest waves in each case.

Oceanography | Vol.31, No.298

REFERENCESAarnes, O.J., M. Reistad, Ø. Breivik, E. Bitner-

Gregersen, L.I. Eide, O. Gramstad, A.K. Magnusson, B. Natvig, and E. Venem. 2017. Projected changes in significant wave height toward the end of the 21st century: Northeast Atlantic. Journal of Geophysical Research 122:3,394–3,403, https://doi.org/ 10.1002/ 2016JC012521.

Aguirre, C., J.A. Rutllant, and M. Falvey. 2017. Wind waves climatology of the south-east Pacific Ocean. International Journal of Climatology 37:4,288–4,301, https://doi.org/ 10.1002/joc.5084.

Aijaz, S., M. Ghantous, A.V. Babanin, I. Ginis, B. Thomas, and G. Wake. 2017. Nonbreaking wave-induced mixing in upper ocean during tropical cyclones using coupled hurricane- ocean- wave modeling. Journal of Geophysical Research 122:3,939–3,963, https://doi.org/ 10.1002/2016JC012219.

Babanin, A.V., and W.E. Rogers. 2014. Generation and limiters of rogue waves. International Journal of Ocean and Climate Systems 5:38–49, https://doi.org/ 10.1260/ 1759-3131.5.2.39.

Barnard, P.L., D. Hoover, D.M. Hubbard, A. Snyder, B.C. Ludka, J. Allan, G.M. Kaminsky, P. Rugggiero, T.W. Gallien, L. Gable, and others. 2017. Extreme oceanographic forcing and coastal response due to the 2015–2016 El Niño. Nature Communications 8:14365, https://doi.org/10.1038/ncomms14365.

Camus, P., I.J. Losada, C. Izaguirre, A. Espejo, M. Menéndez, and J. Pérez. 2017. Statistical wave climate projections for coastal impact assess-ments. Earth’s Future 5:918–933, https://doi.org/ 10.1002/2017EF000609.

Camus, P., M. Menéndez, F.J. Méndez, C. Izaguirre, A. Espejo, V. Cánovas, J. Pérez, A. Rueda, I.J. Losada, and R. Medina. 2014. A weather- type statistical downscaling framework for ocean wave climate. Journal of Geophysical Research 119:7,389–7,405, https://doi.org/ 10.1002/2014JC010141.

Casas-Prat, M., X. Wang, and N. Swart. 2018. CMIP5-based global wave climate projec-tions including the entire Arctic Ocean. Ocean Modelling 123:66–85, https://doi.org/10.1016/ j.ocemod.2017.12.003.

Chen, S.S., and M. Curcic. 2016. Ocean surface waves in Hurricane Ike (2008) and Superstorm Sandy (2012): Coupled model predictions and observa-tions. Ocean Modelling 103:161–176, https://doi.org/ 10.1016/j.ocemod.2015.08.005.

Chen, Y., and X. Yu. 2017. Sensitivity of storm wave modeling to wind stress evaluation meth-ods. Journal of Advances in Modeling Earth Systems 9:893–907, https://doi.org/ 10.1002/ 2016MS000850.

Chen, Y., F. Zhang, B.W. Green, and X. Yu. 2018. Impacts of ocean cooling and reduced wind drag on Hurricane Katrina (2005) based on numerical simulations. Monthly Weather Review 146:287–306, https://doi.org/10.1175/MWR-D-17-0170.1.

Coles, S. 2001. An Introduction to Statistical Modeling of Extreme Values. Springer, 209 pp., https://doi.org/ 10.1007/978-1-4471-3675-0.

Davison, A., and R. Huser. 2015. Statistics of extremes. Annual Review of Statistics and Its Application 2:203–235, https://doi.org/10.1146/annurev-statistics-010814-020133.

Doble, M.J., and J.-R. Bidlot. 2013. Wave buoy mea-surements at the Antarctic sea ice edge com-pared with an enhanced ECMWF WAM: Progress towards global waves-in-ice modelling. Ocean Modelling 70:166–173, https://doi.org/10.1016/ j.ocemod.2013.05.012.

Duvat, V.K.E., A.K. Magnan, S. Etienne, C. Salmon, and C. Pignon-Mussaud. 2016. Assessing the impacts of and resilience to Tropical Cyclone

Bejisa, Reunion Island (Indian Ocean). Natural Hazards 83:601–640, https://doi.org/10.1007/s11069-016-2338-5.

Emanuel, K. 2005. Increasing destructiveness of tropical cyclones over the past 30 years. Nature 436:686–688, https://doi.org/10.1038/nature03906.

Emanuel, K., S. Solomon, D. Folini, S. Davis, and C. Cagnazzo. 2013. Influence of tropical tropo-pause layer cooling on Atlantic hurricane activity. Journal of Climate 26:2,288–2,301, https://doi.org/ 10.1175/JCLI-D-12-00242.1.

Erikson, L., C. Hegermiller, P. Barnard, P. Ruggiero, and M. van Ormondt. 2015. Projected wave condi-tions in the Eastern North Pacific under the influ-ence of two CMIP5 climate scenarios. Ocean Modelling 96:171–185, https://doi.org/10.1016/ j.ocemod.2015.07.004.

Fan, Y., I. Ginis, and T. Hara. 2009. The effect of wind-wave-current interaction on air-sea momentum fluxes and ocean response in tropical cyclones. Journal of Physical Oceanography 39(4):1,019–1,034, https://doi.org/ 10.1175/2008JPO4066.1.

Fan, Y., I.M. Held, S.-J. Lin, and X.L. Wang. 2013. Ocean warming effect on surface gravity wave cli-mate change for the end of the twenty-first century. Journal of Climate 26:6,046–6,066, https://doi.org/ 10.1175/JCLI-D-12-00410.1.

Fan, Y., and W.E. Rogers. 2016. Drag coefficient com-parisons between observed and model simulated directional wave spectra under hurricane condi-tions. Ocean Modelling 102:1–13, https://doi.org/ 10.1016/j.ocemod.2016.04.004.

Harley, M.D., I.L. Turner, M.A. Kinsela, J.H. Middleton, P.J. Mumford, K.D. Splinter, M.S. Phillips, J.A. Simmons, D.J. Hanslow, and A.D. Short. 2017. Extreme coastal erosion enhanced by anoma-lous extratropical storm wave direction. Nature Scientific Reports 7:6033, https://doi.org/10.1038/s41598-017-05792-1.

Hemer, M.A., Y. Fan, N. Mori, A. Semedo, and X.L. Wang. 2013a. Projected changes in wave climate from a multi-model ensemble. Nature Climate Change 3:471–476, https://doi.org/10.1038/nclimate1791.

Hemer, M.A., J. Katzfey, and C.E. Trenham. 2013b. Global dynamical projections of surface ocean wave climate for a future high greenhouse gas emission scenario. Ocean Modelling 70:221–245, https://doi.org/10.1016/j.ocemod.2012.09.008.

Hemer, M.A., X.L. Wang, R. Weisse, and V.R. Swail. 2012. Advancing wind-waves climate science: The COWCLIP project. Bulletin of the American Meteorological Society 93(6)791–796, https://doi.org/ 10.1175/ BAMS-D-11-00184.1.

Hemer, M.A., S. Zieger, T. Durrant, J. O’Grady, R.K. Hoeke, K.L. McInnes, and U. Rosebrock. 2017. A revised assessment of Australia’s national wave energy resource. Renewable Energy 114(Part A):85–107, https://doi.org/10.1016/ j.renene.2016.08.039.

Hoeke, R.K., K.L. McInnes, and J.G. O’Grady. 2015. Wind and wave setup contributions to extreme sea levels at a tropical high island: A stochastic cyclone simulation study for Apia, Samoa. Journal of Marine Science and Engineering 3:1,117–1,135, https://doi.org/ 10.3390/jmse3031117.

Huang, P., I.-I. Lin, C. Chou, and R.-H. Huang. 2015. Change in ocean subsurface environment to suppress tropical cyclone intensification under global warming. Nature Communications 6:7188, https://doi.org/ 10.1038/ncomms8188.

Hurrell, J.W., J.J. Hack, D. Shea, J.M. Caron, and J. Rosinski. 2008. A new sea surface tem-perature and sea ice boundary dataset for the Community Atmosphere Model. Journal of Climate 21:5,145–5,153, https://doi.org/ 10.1175/ 2008JCLI2292.1.

Hurrell, J.W., M.M. Holland, P.R. Gent, S. Ghan, J.E. Kay, P.J. Kushner, J.-F. Lamarque, W.G. Large, D. Lawrence, K. Lindsay, and others 2013. The Community Earth System Model: A frame-work for collaborative research. Bulletin of the American Meteorological Society 94:1,339–1,360, https://doi.org/ 10.1175/ BAMS-D-12-00121.1.

IPCC. 2014. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. C.B Field, V.R. Barros, D.J. Dokken, K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova, B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White, eds, Cambridge University Press, Cambridge, UK, and New York, NY, USA, 1,132 pp.

Izaguirre, C., F.J. Méndez, M. Menéndez, and I.J. Losada. 2011. Global extreme wave height variability based on satellite data. Geophysical Research Letters 38(10), https://doi.org/ 10.1029/ 2011GL047302.

Komen, G.J., L. Cavaleri, M. Donelan, K. Hasselmann, S. Hasselmann, and P.A.E.M. Janssen. 1994. Dynamics and Modelling of Ocean Waves. Cambridge University Press, 532 pp.

Li, H., and R.L. Sriver. 2018. Tropical cyclone activity in the high-resolution Community Earth System Model and the impact of ocean coupling. Journal of Advances in Modeling Earth Systems 10:165–186, https://doi.org/10.1002/2017MS001199.

Lin, I.-I., I.-F. Pun, and C.-C. Lien. 2014. “Category-6” supertyphoon Haiyan in global warming hiatus: Contribution from subsurface ocean warming. Geophysical Research Letters 41:8,547–8,553, https://doi.org/10.1002/2014GL061281.

Mackay, E.B., A.S. Bahaj, and P.G. Challenor. 2010. Uncertainty in wave energy resource assessment: Part 1. Historic data. Renewable Energy 35:1,792–1,808, https://doi.org/10.1016/ j.renene.2009.10.026.

Masselink, G., B. Castelle, T. Scott, G. Dodet, S. Suanez, D. Jackson, and F. Floc’h. 2016. Extreme wave activity during 2013/2014 winter and morpho-logical impacts along the Atlantic coast of Europe. Geophysical Research Letters 43:2,135–2,143, https://doi.org/10.1002/2015GL067492.

Massom, R.A., T.A. Scambos, L.G. Bennetts, P. Reid, V.A. Squire, and S.E. Stammerjohn. 2018. Antarctic ice shelf disintegration triggered by sea ice loss and ocean swell. Nature 558:383–389, https://doi.org/ 10.1038/s41586-018-0212-1.

Méndez, F.J., M. Menéndez, A. Luceno, and I.J. Losada. 2006. Estimation of the long-term vari-ability of extreme significant wave height using a time-dependent peak over threshold (POT) model. Journal of Geophysical Research 111(C7), https://doi.org/ 10.1029/ 2005JC003344.

Mitchell, D., K. AchutaRao, M. Allen, I. Bethke, U. Beyerle, A. Ciavarella, P.M. Forster, J. Fuglestvedt, N. Gillett, K. Haustein, and others. 2017. Half a degree additional warming, progno-sis and projected impacts (HAPPI): Background and experimental design. Geoscientific Model Development 10:571–583, https://doi.org/10.5194/gmd-10-571-2017.

Mizuta, R., A. Murata, M. Ishii, H. Shiogama, K. Hibino, N. Mori, O. Arakawa, Y. Imada, K. Yoshida, T. Aoyagi, and others. 2017. Over 5,000 years of ensemble future climate simulations by 60-km global and 20-km regional atmospheric mod-els. Bulletin of the American Meteorological Society 98(7):1,383–1,398, https://doi.org/10.1175/BAMS-D-16-0099.1.

Moon, I.-J., I. Ginis, T. Hara, H.L. Tolman, C.W. Wright, and E.J. Walsh. 2003. Numerical simulation of sea surface directional wave spectra under hurricane wind forcing. Journal of Physical Oceanography 33:1,680–1,706, https://doi.org/ 10.1175/2410.1.

Oceanography | June 2018 99

Munich RE. 2018. Natural catastrophe review: Series of hurricanes makes 2017 year of highest insured losses ever. Press release, January 4, 2018, https://www.munichre.com/en/media-relations/ publications/press-releases/2018/2018-01-04-press-release/index.html.

Murakami, H., R. Mizuta, and E. Shindo. 2012. Future changes in tropical cyclone activity projected by multi-physics and multi-SST ensemble experi-ments using the 60-km-mesh MRI-AGCM. Climate Dynamics 39:2,569–2,584, https://doi.org/10.1007/s00382-011-1223-x.

Murakami, H., G.A. Vecchi, T.L. Delworth, A.T. Wittenberg, S. Underwood, R. Gudgel, X. Yang, L. Jia, F. Zeng, K. Paffendorf, and W. Zhang. 2017. Dominant role of subtropical Pacific warming in extreme eastern Pacific hurricane seasons: 2015 and the future. Journal of Climate 30:243–264, https://doi.org/10.1175/JCLI-D-16-0424.1.

Murakami, H., G.A. Vecchi, S. Underwood, T.L. Delworth, A.T. Wittenberg, W.G. Anderson, J.-H. Chen, R.G. Gudgel, L.M. Harris, S.-J. Lin, and F. Zeng. 2015. Simulation and prediction of cat-egory 4 and 5 hurricanes in the high-resolution GFDL HiFLOR coupled climate model. Journal of Climate 28:9,058–9,079, https://doi.org/10.1175/JCLI-D-15-0216.1.

NOAA NCEI (National Centers for Environmental Information). 2018. U.S. Billion-Dollar Weather and Climate Disasters, https://www.ncdc.noaa.gov/billions/.

Patricola, C.M., P. Chang, and R. Saravanan. 2016. Degree of simulated suppression of Atlantic trop-ical cyclones modulated by flavour of El Niño. Nature Geoscience 9:155–160, https://doi.org/ 10.1038/NGEO2624.

Patricola, C.M., R. Saravanan, and P. Chang. 2014. The Impact of the El Niño-Southern Oscillation and Atlantic Meridional Mode on seasonal Atlantic tropical cyclone activity. Journal of Climate 27:5,311–5,328, https://doi.org/10.1175/JCLI-D-13-00687.1.

Patricola, C., and M. Wehner. In press. Anthropogenic influences on major tropical cyclone events. Nature.

Perez, J., M. Menéndez, P. Camus, F.J. Méndez, and I.J. Losada. 2015. Statistical multi-model climate projections of surface ocean waves in Europe. Ocean Modelling 96:161–170, https://doi.org/ 10.1016/ j.ocemod.2015.06.001.

Risser, M.D., and M.F. Wehner. 2017. Attributable human-induced changes in the likelihood and magnitude of the observed extreme precipi-tation during Hurricane Harvey. Geophysical Research Letters 44(2):12,457–12,464, https://doi.org/10.1002/2017GL075888.

Ruggiero, P., P.D. Komar, and J.C. Allan. 2010. Increasing wave heights and extreme value pro-jections: The wave climate of the U.S. Pacific Northwest. Coastal Engineering 57:539–552, https://doi.org/10.1016/j.coastaleng.2009.12.005.

Seroka, G., T. Miles, Y. Xu, J. Kohut, O. Schofield, and S. Glenn. 2016. Hurricane Irene sensitivity to strat-ified coastal ocean cooling. Monthly Weather Review 144:3,507–3,530, https://doi.org/10.1175/MWR-D-15-0452.1.

Shimura, T., N. Mori, and M.A. Hemer. 2016. Variability and future decreases in winter wave heights in the Western North Pacific. Geophysical Research Letters 43:2,716–2,722, https://doi.org/ 10.1002/2016GL067924.

Shimura, T., N. Mori, and M.A. Hemer. 2017. Projection of tropical cyclone-generated extreme wave cli-mate based on CMIP5 multi-model ensem-ble in the Western North Pacific. Climate Dynamics 49:1,449–1,462, https://doi.org/10.1007/s00382-016-3390-2.

Shimura, T., N. Mori, and H. Mase. 2015. Future projections of extreme ocean wave climates and the relation to tropical cyclones: Ensemble

experiments of MRI-AGCM3.2H. Journal of Climate 28:9,838–9,856, https://doi.org/10.1175/JCLI-D-14-00711.1.

Shope, J.B., C.D. Storlazzi, L.H. Erikson, and C.A. Hegermiller. 2016. Changes to extreme wave climates of islands within the Western Tropical Pacific throughout the 21st century under RCP 4.5 and RCP 8.5, with implications for island vulner-ability and sustainability. Global and Planetary Change 141:25–38, https://doi.org/10.1016/ j.gloplacha.2016.03.009.

Skamarock, W.C., J.B. Klemp, J. Dudhia, D.O. Gill, D.M. Barker, M.G. Duda, X.-Y. Huang, W. Wang, and J.G. Powers. 2008. A Description of the Advanced Research WRF Version 3. NCAR Technical Note NCAR/TN-475+STR, 125 pp., https://doi.org/ 10.5065/D68S4MVH.

Sobel, A.H., S.J. Camargo, T.M. Hall, C.-Y. Lee, M.K. Tippett, and A.A. Wing. 2016. Human influence on tropical cyclone intensity. Science 353:242–246, https://doi.org/10.1126/ science.aaf6574.

Sriver, R.L. 2016. Observational evidence supports the role of tropical cyclones in regulating climate. Proceedings of the National Academy of Sciences of the United States of America 38:15,173–15,174, https://doi.org/10.1073/pnas.1314721110.

Taylor, K.E., R.J. Stouffer, and G.A. Meehl. 2012. An overview of CMIP5 and the experiment design. Bulletin of the American Meteorological Society 93:485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.

Timmermans, B., D. Stone, M. Wehner, and H. Krishnan. 2017. Impact of tropical cyclones on modeled extreme wind-wave climate. Geophysical Research Letters 44:1,393–1,401, https://doi.org/ 10.1002/2016GL071681.

Tolman, H.L., and the WAVEWATCH III Development Group. 2014. User Manual and System Documentation of WAVEWATCH III, version 4.18. Environmental Modeling Center Marine Modeling and Analysis Branch, National Oceanic and Atmospheric Administration, 282 pp. plus appendices.

Vecchi, G.A., and B.J. Soden. 2007a. Effect of remote sea surface temperature change on tropical cyclone potential intensity. Nature 450:1,066–1,071, https://doi.org/10.1038/nature06423.

Vecchi, G.A., and B.J. Soden. 2007b. Increased tropical Atlantic wind shear in model pro-jections of global warming. Geophysical Research Letters 34(8), https://doi.org/ 10.1029/ 2006GL028905.

Wang, D.W., D.A. Mitchell, W.J. Teague, E. Jarosz, and M.S. Hulbert. 2005. Extreme waves under Hurricane Ivan. Science 309:896, https://doi.org/ 10.1126/science.1112509.

Wang, X.L., Y. Feng, and V.R. Swail. 2014. Changes in global ocean wave heights as projected using multimodel CMIP5 simulations. Geophysical Research Letters 41:1,026–1,034, https://doi.org/ 10.1002/2013GL058650.

Wang, X.L., A. Semedo, M. Hemer, M. Dobrynin, and COWCLIP contributors. 2016. Report of the 2016 Meeting for the WCRP-JCOMM Coordinated Ocean Wave Climate Project (COWCLIP). Technical Report, Joint WMO/IOC Technical Commission for Oceanography and Marine Meteorology (JCOMM), 35 pp.

Wang, X.L., and V.R. Swail. 2006. Climate change signal and uncertainty in projections of ocean wave heights. Climate Dynamics 26:109–126, https://doi.org/ 10.1007/s00382-005-0080-x.

Wehner, M., Prabhat, K.A. Reed, D. Stone, W.D. Collins, and J. Bacmeister. 2015. Resolution dependence of future tropical cyclone projections of CAM5.1 in the U.S. CLIVAR hurricane working group idealized configurations. Journal of Climate 28:3,905–3,925, https://doi.org/10.1175/JCLI-D-14-00311.1.

Young, I. 1999. Seasonal variability of the global ocean wind and wave climate. International Journal of Climatology 19:931–950, https://doi.org/ 10.1002/(SICI)1097-0088(199907)19:9<931::AID-JOC412>3.0.CO;2-O.

Young, I.R., J. Vinoth, S. Zieger, and A.V. Babanin. 2012. Investigation of trends in extreme value wave height and wind speed. Journal of Geophysical Research 117(C11), https://doi.org/ 10.1029/2011JC007753.

Zambon, J.B., R. He, and J.C. Warner. 2014. Investigation of Hurricane Ivan using the coupled ocean- atmosphere- wave- sediment transport (COASWT) model. Ocean Dynamics 64:1,535–1,554, https://doi.org/10.1007/s10236-014-0777-7.

Zhang, W., G.A. Vecchi, H. Murakami, T. Delworth, A.T. Wittenberg, A. Rosati, S. Underwood, W. Anderson, L. Harris, R. Gudgel, and others. 2016. Improved simulation of tropical cyclone responses to ENSO in the Western North Pacific in the high-resolution GFDL HiFLOR coupled cli-mate model. Journal of Climate 29:1,391–1,415, https://doi.org/ 10.1175/ JCLI-D-15-0475.1.

Zuidema, P., P. Chang, B. Medeiros, B.P. Kirtman, R. Mechoso, E.K. Schneider, T. Toniazzo, I. Richter, R.J. Small, K. Bellomo, and others. 2016. Challenges and prospects for reducing coupled climate model SST biases in the east-ern tropical Atlantic and Pacific Oceans: The U.S. CLIVAR eastern tropical oceans synthesis work-ing group. Bulletin of the American Meteorological Society 97(12):2,305–2,327, https://doi.org/10.1175/BAMS-D-15-00274.1.

ACKNOWLEDGMENTSThis material is based upon work supported by the Regional and Global Climate Modeling Program of the US Department of Energy, Office of Science, Office of Biological and Environmental Research, under contract number DE-AC02-05CH11231. These simulations were performed using resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility sup-ported by the Office of Science of the US Department of Energy, also under contract no. DE-AC02-05CH11231. All data buoy observations were obtained from the US National Oceanic and Atmospheric Administration (NOAA) National Data Buoy Center (NDBC) at http://www.ndbc.noaa.gov. We thank two anonymous reviewers for comments that helped improve this paper.

AUTHORSBen Timmermans (ben.timmermans@gmail.com) is Postdoctoral Fellow and Christina Patricola is Research Scientist, Climate and Ecosystems Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA. Michael Wehner is Senior Staff Scientist, Computational Research Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA.

ARTICLE CITATIONTimmermans, B., C. Patricola, and M. Wehner. 2018. Simulation and analysis of hurricane-driven extreme wave climate under two ocean warming scenarios. Oceanography 31(2):88–99, https://doi.org/10.5670/oceanog.2018.218.